This paper aims to present a brief overview of the development of sensor fusion in various application in recent years, and to understand the challenges and ability of sensor fusion. Various algorithms that are typically employed are covered to comprehend the complexity of usage in different scenarios.

Sensor fusion: A review of methods and applications

Integrating new candidate sensors, such as Global Navigation Satellite System (GNSS) and inertial measurement unit (IMU), into fail-safe train positioning systems have recently become a prominent area of research. Although there are a number of contributions related to the design of data fusion algorithms, the lack of details in raw measurements analysis has directly motivated this paper. This paper aims to record data from a variety of sensors (such as GNSS, IMU, magnetometer, barometer, tachometers, and Doppler radars) to evaluate train velocity and railway features (track slope, curve cant, and radius) extending previous works in the instrumentation and measurement field. The field test designed and concisely described in this paper presents several challenging environments, such as a tunnel, which can be used to analyze the candidate sensors limitations. In addition, a demonstration of a data fusion algorithm is presented to calculate train velocity based on measurements from the candidate sensors. The results obtained by an extended Kalman filter using GNSS and IMU are compared with velocity recorded by tachometers and Doppler radars, which is considered to be the reference value. The calculated velocity by IMU and GNSS when both sensors measurements are available presents an absolute error in velocity lower than 2 km/h in more than 90% of test duration. Finally, railway features (curve radius, cant, and slope) are calculated and analyzed according to train and railway dynamics.

Evaluation of Experimental GNSS and 10-DOF MEMS IMU Measurements for Train Positioning

In this paper, a new control scheme for motion control of quadrotors is presented. The proposed model based controller makes use of simple estimations of vehicle parameters. Besides, the controller is developed under properties of the vehicle model. A rigorous analysis is developed to prove that the closed-loop system trajectories are uniformly ultimately bounded. The validity of the proposed control scheme is tested by numerical simulations and experimental results. This validation shows that the proposed controller guarantees that the quadrotor performs tracking of a desired pose (position and orientation) trajectory with good accuracy. The new scheme has been compared with respect to other robust controllers. The results indicate better tracking accuracy for the new scheme.

A robust approach for trajectory tracking control of a quadrotor with experimental validation.

These days, the accuracy of the measured tilt angle is extremely important and it is always a challenge especially for robotics application. MEMS (micro-electro-mechanical system) gyroscope and accelerometer has been widely used to measure the tilt angle along the axis. However, various studies highlighted that MEMS gyroscope and accelerometer has weakness to gain accuracy of orientation. In order to cater the low accuracy, data from the MEMS gyroscope and accelerometer require a filter where the data from both sensors will be combined by using the complementary filter. Thus, digital complementary filter is developed in such a way that the strength of one sensor will be used to overcome the weaknesses of the other sensor which is complementary to each other. The filtering will be demonstrated using MEMS gyroscope and accelerometer in different position throughout the measurement for analysis and angle estimation via complementary filter. From the simulation result, it is evident that the complementary filter method is achieved with responsive and accurate angle of 0 to 90 degrees with same as actual position at factor value of 0.97. In addition, it is not sensitive to horizontal accelerations or to gyroscope drift with microcontroller-friendly filtering system.

Low cost MEMS gyroscope and accelerometer implementation without Kalman Filter for angle estimation

A quadrotor helicopter control system using cascade PID is designed in this paper. Aim at dealing with the data drift and mutation caused by quadrotor badly vibration. Kalman filter combines with Moving-average is used to filter out these noise. The quaternion method is applied to calculate the attitude. The real time attitude adjustment is controlled by cascade PID control, which inner loop uses PID control of angular velocity and outer loop uses PI control of angle. Experiment results show that, the filter combination improved the reliability of the accelerometer data. The cascade PID algorithm is efficient for attitude control which meet the response for real-time control in the perturbations condition.

Quadrotor helicopter Attitude Control using cascade PID

This paper describes the design and integration of the instrumentation and sensor fusion that is used to allow the autonomous flight of a quadrotor. A comercial frame is used, a mathematical model for the quadrotor is developed and its parameters determined from the characterization of the unit. A 9 degrees of freedom Inertial Measurement Unit (IMU) equipped with a barometer is calibrated and added to the platform. Sensor fusion is done by two modified Extended Kalman Filters (EKF): one combining data provided by IMU and the other also including the information provided by GPS. A reliable estimation of the state variables is obtained. Three states representing systematic bias in the accelerometer measurements are also added to the EKF, which improves the inertial estimation of the position. A stable autonomous platform is achieved.

Design and implementation of sensor data fusion for an autonomous quadrotor

In this paper we describe the stages needed to achieve an autonomous stable platform with quadrotor architecture. Based on a commercial frame, we first design and integrate the instrumentation. An Inertial Measurement Unit (IMU), which also has a barometer, is added to the platform and accurately calibrated. The sensor fusion and signal filtering are done based on an Extended Kalman Filter (EKF). Finally, a proportional-integral controller based on an enhanced Linear Quadratic Regulator (LQR) algorithm is implemented in the control system, using the mathematical model of the quadrotor. The parameters of this model are obtained from the characterization of the unit, achieving at the end, a stable autonomous platform.

Rodrigo Rosa

Papers

Design and implementation of sensor data fusion for an autonomous quadrotor

Design and implementation of flight control for an autonomous quadrotor